The AWS Worldwide Specialist Organization (WWSO) is a team of go-to-market experts that support strategic services, customer segments, and verticals at AWS. Working together, the Specialist Insights Team (SIT) and the Finance, Analytics, Science, and Technology team (FAST) support WWSO in acquiring, securing, and delivering information and business insights at scale by working with the broader AWS community (Sales, Business Units, Finance) enabling data-driven decisions to be made.

SIT is made up of analysts who bring deep knowledge of the business intelligence (BI) stack to support the WWSO business. Some analysts work across multiple areas, whereas others are deeply knowledgeable in their specific verticals, but all are technically proficient in BI tools and methodologies. The team’s ability to combine technical and operational knowledge, in partnership with domain experts within WWSO, helps us build a common, standard data platform that can be used throughout AWS.

Untapped potential in data availability

One of the ongoing challenges for the team was how to turn the 2.1 PB of data inside the data lake into actionable business intelligence that can drive actions and verifiable outcomes. The resources needed to translate the data, analyze it, and succinctly articulate what the data shows had been a blocker of our ability to be agile and responsive to our customers.

After reviewing several vendor products and evaluating the pros and cons of each, Amazon QuickSight was chosen to replace our existing legacy BI solution. It not only satisfied all of the criteria necessary to provide actionable insights across WWSO business but allows us to scale securely across tens of thousands of users at AWS.

In this post, we discuss what influenced the decision to implement QuickSight, and will detail some of the benefits our team has seen since implementation.

Legacy tool deprecation

The legacy BI solution presented a number of challenges, starting with scaling, complex governance, and siloed reporting. This resulted in poor performance, cumbersome development processes, multiple versions of truth, and high costs. Ultimately, the legacy BI solution had significant barriers to widespread adoption, including long time to insights, lack of trust, low innovation, and return-on-investment (ROI) justification.

After the decision was made to deprecate the previous BI tool our team had been using to provide reports and insights to the organization, the team began to make preparations for the impending switch. We met with analysts across the specialist organization to gather feedback on what they’d like to see in the next iteration of reporting capabilities. Based on that feedback, and with guidance from our leadership teams, the following criteria needed to be met in our next BI tool:

  • Accessible insights – To ensure users with varying levels of technical aptitude could understand the information, the insights format needed to be easy to understand.
  • Speed – With millions of records, processing speed needed to be lightning fast, and we also didn’t want to invest a lot of time in technical implementation or user education training.
  • Cost – Being a frugal company, we needed to ensure that our BI solution would not only do what we needed it to do but that it wouldn’t blow up our budget.
  • Security – Built-in row-level security, and a custom solution developed internally, had the ability to give access to thousands of users across AWS.

Among other considerations that ultimately influenced the decision to use QuickSight was that it’s a fully managed service, which meant no need to maintain a separate server or manage any upgrades. Because our team handles sensitive data, security was also top of mind. QuickSight passed that test as well; we were able to implement fine-grained security measures and saw no trade-off in performance.

A simple, speedy, single source of truth

With such a wide variety of teams needing access to the data and insights our team provides, our BI solution needed to be user-friendly and intuitive without the need for extensive training or convoluted instructions. With millions of records used to generate insights on sales pipelines, revenue, headcount, etc., queries could become quite complex. To meet our first top priority for accessible insights, we were looking for a combination of easy-to-operate and easy-to-understand visualizations.

Once our QuickSight implementation was complete, near-real-time, actionable insights with informative visuals were just a few clicks away. We were impressed by how simple it was to get at-a-glance insights that told data-driven stories about the key performance indicators that were most important to our stakeholder community. For business-critical metrics, we’re able to set up alerts that trigger emails to owners when certain thresholds are met, providing peace of mind that nothing important will slip through the cracks.

With the goal of migrating 400+ dashboards from the legacy BI solution over to QuickSight successfully, there were three critical components that we had to get right. Not only did we need to have the right technology, we also needed to set up the right processes while also keeping change management—from a people perspective—top of mind.

This migration project provided us with an opportunity to standardize our data, ensuring that we have a uniform source of truth that enables efficiency, as well as governed access and self-service across the company. In the spirit of working smarter (not harder), we kickstarted the migration in parallel with the data standardization project.

We started by establishing clear organization goals for alignment and a solid plan from start to finish. Next steps were to focus on row-level security design and evolution to ensure that we can provide governance and security at scale. To ensure success, we first piloted migrating 35+ dashboards and 500+ users. We then established a core technical team whose focus was to be experts at QuickSight and migrate another 400+ dashboards, 4,000+ users, and 60,0000+ impressions. The technical team also trained other members of the team to bring everyone along on the change management journey. We were able to complete the migration in 18 months across thousands of users.

With the base in place, we shifted focus to move from foundational business metrics to machine learning (ML) based insights and outcomes to help drive data-driven actions.

The following screenshot shows an example of what one of our QuickSight dashboards looks like, though the numbers do not reflect real values; this is test data.

With speed being next on our list of key criteria, we were delighted to learn more about how QuickSight works. SPICE, an acronym for Super-fast, Parallel, In-memory Calculation Engine, is the robust engine that QuickSight uses to rapidly perform advanced calculations and serve data. The query runtimes and dashboard development speed were both appreciably faster in comparison to other data visualization tools we had used, where we’d need to wait for it to process every time we added a new calculation or a new field to the visual. The dashboard load times were also noticeably faster than the load times from our previous BI tool; most load in under 5 seconds, compared to several minutes with the previous BI tool.

Another benefit of having chosen QuickSight is that there has been a significant reduction in the number of disagreements over data definitions or questions about discrepancies between reports. With standardized SPICE datasets built in QuickSight, we can now offer data as a service to the organization, creating a single source of truth for our insights shared across the organization. This saved the organization hours of time investigating unanswered questions, enabling us to be more agile and responsive, which makes us better able to serve our customers.

Dashboards are just the beginning

We’re very happy with QuickSight’s performance and scalability, and we are very excited to improve and expand on the solid reporting foundation we’ve begun to build. Having driven adoption from 50 percent to 83 percent, as well as seeing a 215 percent growth in views and a 157 percent growth in users since migrating to QuickSight, it’s clear we made the right choice.

We were intrigued by a recent post by Amy Laresch and Kellie Burton, AWS Analytics sales team uses QuickSight Q to save hours creating monthly business reviews. Based on what we learned from that post, we also plan to test out and eventually implement Amazon QuickSight Q, a ML powered natural language capability that gives anyone in the organization the ability to ask business questions in natural language and receive accurate answers with relevant visualizations. We’re also considering integrations with Amazon SageMaker and Amazon Honeycode-built apps for write back.

To learn more, visit Amazon QuickSight.


About the Authors

David Adamson is the head of WWSO Insights. He is leading the team on the journey to a data driven organization that delivers insightful, actionable data products to WWSO and shared in partnership with the broader AWS organization. In his spare time, he likes to travel across the world and can be found in his back garden, weather permitting exploring and photography the night sky.

Yash Agrawal is an Analytics Lead at Amazon Web Services. Yash’s role is to define the analytics roadmap, develop standardized global dashboards & deliver insightful analytics solutions for stakeholders across AWS.

Addis Crooks-Jones is a Sr. Manager of Business Intelligence Engineering at Amazon Web Services Finance, Analytics and Science Team (FAST). She is responsible for partnering with business leaders in the World Wide Specialist Organization to build a culture of data  to support strategic initiatives. The technology solutions developed are used globally to drive decision making across AWS. When not thinking about new plans involving data, she like to be on adventures big and small involving food, art and all the fun people in her life.

Graham Gilvar is an Analytics Lead at Amazon Web Services. He builds and maintains centralized QuickSight dashboards, which enable stakeholders across all WWSO services to interlock and make data driven decisions. In his free time, he enjoys walking his dog, golfing, bowling, and playing hockey.

Shilpa Koogegowda is the Sales Ops Analyst at Amazon Web Services and has been part of the WWSO Insights team for the last two years. Her role involves building standardized metrics, dashboards and data products to provide data and insights to the customers.

Source: https://aws.amazon.com/blogs/big-data/aws-specialist-insights-team-uses-amazon-quicksight-to-provide-operational-insights-across-the-aws-worldwide-specialist-organization/